Sentinel lymph node (SLN) biopsy is a minimally invasive means of accurately staging the axilla in breast cancer patients. While finding negative SLNs avoids axillary lymph node dissection (ALND), the finding of micrometastasis in the SLN mandates a completion ALND. Up to 80% of all SLN-positive patients will have no further disease in the axilla. There is a significant gap in our current knowledge in predicting which SLN- positive patients have a low likelihood of non-SLN metastasis and can therefore avoid ALND. A novel intra- operative RT-PCR based assay (GeneSearch(r)) for detecting SLN micrometastases has been shown to correlate with SLN tumor burden in breast cancer patients. The potential utility of this new technology in predicting which patients will not have further disease in their non-SLNs has not been investigated. Developing a robust clinical prediction rule for non-SLN status using these data will fill this gap in our knowledge and provide timely conclusions, which will have a direct impact on patient care. In order to achieve this over-arching objective, we propose the following specific aims: (1) To determine the impact of quantitative cycle time values of the GeneSearch(r) assay on non-SLN status using both bivariate and multivariable logistic regression analyses, and (2) to create and validate (using bootstrap techniques) a clinical prediction rule to predict non-SLN status in SLN-positive patients using preoperatively- and intraoperatively-available clinicopathologic factors, including data obtained from the GeneSearch(r) assay. We hypothesize that the quantitative RT-PCR cycle time for both mammoglobin (MMG) and cytokeratin 19 (CK19), as evaluated as continuous values using the GeneSearch(r) assay, will be significantly associated with non-SLN metastasis on bivariate analyses. In addition, the cycle time for these markers will predict non-SLN metastasis independent of other factors. We hypothesize that a valid clinical prediction rule can be created using preoperatively- and intraoperatively-available clinicopathologic factors, including data obtained from the GeneSearch(r) assay that will identify a subset of patients in whom the likelihood of non-SLN metastasis is = 5%. It is anticipated that the addition of data from the GeneSearch(r) assay will significantly improve the ability to predict this group of patients over the use of clinicopathologic factors alone. The creation and validation of a robust clinical prediction rule that incorporates novel molecular data to identify a subgroup of patients at low risk of having non-SLN metastasis will fill a significant gap in our current knowledge and will reduce unnecessary morbidity of ALND, helping to meet the NCI's Challenge of eliminating suffering due to cancer by 2015. PUBLIC HEALTH RELEVANCE: Up to 80% of breast cancer patients who have minimal disease in their first draining (sentinel) lymph nodes will have no further disease in their axilla. However, currently there is no accurate means of predicting which patients will not have residual disease, and therefore all patients who have a tumor deposit = 0.2 mm in their sentinel lymph nodes will have the remaining lymph nodes removed -- a procedure that is associated with considerable morbidity. We seek to create and validate a clinical prediction rule, incorporating novel molecular data, to identify a subgroup of patients at low risk of having non-SLN metastasis, thereby filling a significant gap in our current knowledge and reducing unnecessary morbidity for breast cancer patients.